Nitrogen (N) removal from high-strength wastewater can be accomplished in single-stage combined nitritation-anammox reactors with suspended growth biomass composed of floccular sludge, granular sludge, or of any mix of these 2 different sludge fractions. To date, the influence of floccular biomass on granular sludge reactor performance and stability has not been investigated experimentally or numerically. To address this knowledge gap, two 1D multi-species models were developed in Aquasim to assess the importance of small levels of flocs in putatively granular sludge combined nitritation-anammox reactors for different bulk oxygen concentrations and organics loads. The models included the growth and decay of aerobic ammonium-oxidizing organism (AOO), nitrite-oxidizing organisms (NOO), heterotrophic organisms (OHO), and anammox organisms (AMO) in exclusively granular sludge reactors, and in granular sludge reactors with small levels (∼5% of total biomass) of flocs. While maximum N removal efficiencies were similar for both model structures, floc addition led to a lower optimal dissolved oxygen concentration (DO) as well as a narrower maximum N removal peak, suggesting that small levels of floccular material may decrease process robustness to bulk oxygen changes. For some DO levels, this led to drastic efficiency drops. Furthermore, floc addition also led to substantial segregation in activity and microbial population distribution, with AOO, NOO and OHO concentrated in flocs and AMO concentrated in granules. Increased organic loading (COD:N = 4:3) improved maximum N removal efficiency in both model structures, but yielded substantially different predictions for optimal DO setpoint and process robustness to variations in DO. Taken together, our results indicate that even small levels of floccular biomass in biofilm reactors can have profound implications for reactor performance and optimization and for segregation of linked microbial processes, and suggest that the common practice of neglecting small levels of floccular material in biofilm models and in practice may lead to erroneous predictions.